Abstract Improving diets through increased food and vegetable (F&V) consumption significantly reduces the risk of car- diovascular disease (CVD). Programs increasing the accessibility and affordability of F&Vs among low-income Americans have been hindered by the food consumption cycle associated with poverty: the tendency to over-consume calories shortly after receiving funds at the beginning of each month, draining the budget for F&V purchases, or for all food purchases, by month's end. Increasing food assistance funding alone does not resolve this problem, but an emerging theory about dietary behavior suggests that providing funds for food in smaller installments distributed throughout the month will smooth the consumption cycle and improve healthy eating?counteracting the tendency to respond to lump sum, once-monthly funding installments by purchasing calorie-dense foods immediately after funds are received. The theory also suggests that funds tar- geted toward specific healthy foods (e.g., F&Vs) will improve diets more than untargeted funds, despite the in- convenience of utilizing targeted funds. Our preliminary data support both hypotheses, which we will rigor- ously test in a real-world setting by comparing alternative approaches for delivering food purchasing vouchers. We have established and tested the infrastructure to provide vouchers accepted by all major food stores (e.g., supermarkets, corner shops) in two low-income neighborhoods. Leveraging this innovative infrastructure, in Aim 1 we propose a randomized trial with a two-by-two factorial design, comparing $20 of vouchers valid for one month to four $5 vouchers each valid for a sequential week of the month (lump sum versus distributed funding), and comparing vouchers restricted to F&V purchases to vouchers redeemable for any food (targeted versus untargeted funding). Low-income adults (N=288) recruited through our community partners will be ran- domized to one of four 6-month interventions: monthly targeted, monthly untargeted, weekly targeted, or week- ly untargeted vouchers. Participants will be assessed through efficient verbal 24-hour dietary recalls validated among low-literacy populations, to determine daily consumption of F&Vs and metrics of overall dietary quality at months 0, 6 and 12 (6 months after vouchers end). Additional surveys will identify moderators and mediators of dietary improvement. In Aim 2, we will calculate the cost-effectiveness of the voucher programs. Vouchers and other interventions that support healthy diets reduce population-level CVD disparities over the course of decades. Hence, the full costs and benefits of such interventions are not possible to directly observe through time-limited trials. Following recent NIH policy statements, we will overcome this problem by applying an inno- vative, validated microsimulation modeling approach?a systems science method that integrates trial results with comprehensive data on the effect of dietary changes on CVD risks and costs over the life course. We will identify which voucher delivery strategies cost-effectively reduce CVD disparities?addressing a key scientific uncertainty in healthy eating intervention research.